Skip to content
DT Nova

How much does custom software development cost? What sets the budget

How much does custom software development cost? The honest answer is that custom software development cost is set by scope: how many kinds of users the system serves, how many workflows it automates, which other systems it talks to, how much data moves into it, what security and compliance it must meet, and how it is run after launch. This guide explains each of those drivers, the costs that continue after launch, how to compare building with buying, and how pricing models change the risk, with public sources, so you can compare quotes on the same basis.

This guide does not quote our own prices. It explains how the cost of custom software is built, so any quote you receive, from us or from anyone else, makes sense. For mobile apps specifically, see our guide to app development cost. If the software is the business itself, see our guide on how to start a software company. For AI projects, which add usage-based running costs, see artificial intelligence cost. For specific systems and projects, see our guides to CRM development cost, ERP implementation cost, inventory management software cost, data migration cost, API integration cost and software maintenance cost. For web-based systems, see web app development cost, client portal cost, intranet cost, document management system cost and booking system cost.

The short version

Every custom software budget is the sum of four parts:

  1. The first version: discovery, design, building, testing, data migration and launch. People's time is almost all of it.
  2. Running costs: hosting, databases, third-party services and licences, which grow with use.
  3. Maintenance: security updates, upgrades, fixes and monitoring, every year the system is in use.
  4. Change: new features and adjustments as the business changes, which is the point of owning the software.

The biggest cost: people's time

Custom software is made of hours: understanding the work, deciding what to build, designing it, writing and reviewing code, testing, moving data, training users and managing the project. What an hour costs depends largely on where the team works. Eurostat's release of 31 March 2026 (EU hourly labour costs ranged from €12 to €57 in 2025, read on 11 November 2026) puts average hourly labour costs across the whole economy at €34.9 in the EU and €38.2 in the euro area, from €12.0 in Bulgaria to €56.8 in Luxembourg. Those are labour cost figures for the whole economy, not software rates, but they explain why the same scope can be quoted several times higher in one country than another.

A lower rate does not always mean a lower total. Time-zone distance, unclear ownership of decisions, rework and management overhead all add hours, and a team that has built similar systems before needs fewer of them. Compare the total for the same scope, and the risk that the total grows, rather than hourly rates.

What drives the hours

DriverWhy it adds workHow to keep it in check
User rolesEach role has its own screens, permissions and testsStart with the roles that create the value; add the rest later
WorkflowsEach workflow needs rules, states, exceptions and error handlingAutomate the frequent cases; leave rare exceptions manual at first
IntegrationsEvery outside system has its own interface, limits and failure modesIntegrate only what the first version needs; prefer documented APIs
Data migrationOld data is incomplete, duplicated and inconsistent, and must be cleaned and mappedDecide early which history really needs to move
ReportingReports need data models, filters, exports and access controlLaunch with the reports used weekly; use a BI tool for the rest
Security and complianceSensitive data, audit trails, access reviews and regulated processes add design and testingCollect only what is needed; build on proven identity and hosting services
AvailabilitySystems that must not stop need redundancy, monitoring and on-duty supportState the real requirement: office hours, or around the clock
InterfacesWeb, mobile, offline use and several languages each add building and testingChoose the interfaces users actually need first
Unclear scopeDecisions made late cause rework, which is the most expensive kind of hourPay for discovery before the build; write the scope down

Three common shapes of custom software

  • An internal tool. One team, a handful of workflows, data from one or two existing systems, used during office hours. Replacing a critical spreadsheet or a manual process is usually this shape. See internal tools development.
  • A business system. Several roles and departments, many integrations, data migration from an old system, reporting and audit requirements. CRM, ERP extensions, scheduling and field service systems are this shape, and integrations and migration often take as much work as the screens. See how a custom software development company prices this kind of system in fixed phases.
  • A software product. Sold to many customers, so it needs multi-tenancy, sign-up, billing, onboarding, support tools and higher security and availability. It is the most demanding shape, and it is never finished. See startup software development.

Costs after launch

Software is never finished, because everything around it moves. Operating systems, browsers, libraries and the services it connects to change, security vulnerabilities are published, and the business itself changes. Maintenance is usually described in four kinds: corrective (fixing faults), adaptive (keeping up with a changing environment), perfective (improving what works) and preventive (reducing future problems). Plan a yearly budget for all four, plus hosting and third-party services that grow with use, monitoring, backups and support for users.

Two practical rules keep these costs predictable. First, agree a written maintenance scope before launch: what is covered, response times and what counts as new work. Second, keep the system on mainstream, well-supported technology, so that maintenance never depends on one team or one person.

Build or buy: compare the total

Buying a product is often the right choice, and a good supplier will say so. Ready-made software wins when your process is standard, the product covers it, and the subscription over the system's life costs less than building and running your own. Custom software wins when the process is what makes you different, when you would otherwise run a product plus spreadsheets around its gaps, or when per-user subscriptions grow faster than the business. Our build vs buy software guide gives a weighted scoring grid for the decision.

Compare the total over five years, for the same users and scope: subscriptions or licences, implementation and configuration, integrations, the staff time spent on workarounds, and the cost of leaving. Leaving is becoming cheaper in the EU: under the Data Act, from 12 January 2027 providers of data processing services may not impose switching charges for the switching process, and until then may impose only reduced charges that do not exceed their direct costs (Regulation (EU) 2023/2854, Article 29, read on 11 November 2026). With custom software, the code is a one-time cost you own; the comparison is with a cost you rent.

Pricing models and who carries the risk

The table sums up the main models; our fixed price vs time and materials guide covers the choice in depth.

ModelHow it worksWho carries the risk
Fixed priceAn agreed price for a written scopeThe supplier carries the risk of underestimating; the buyer carries the risk of a scope that was wrong
Time and materialsHours billed as they are worked, often against an estimateThe buyer carries the risk of overruns; flexibility is highest
Phased fixed priceA fixed price for each phase, the next priced on what the last one learnedShared: each commitment is small and based on evidence
Dedicated teamA team reserved for a monthly fee, working through a backlog you prioritiseThe buyer carries the scope risk; suits long-running products

Fixed price is the safer choice for most buyers when the scope can be written down. When it cannot yet, pay for a short, fixed-price discovery phase that writes it, then fix the price of the build. Our guide to software development methodologies explains how these models fit waterfall and agile ways of working.

Cost estimation for software development: how estimates are made

Cost estimation for software development is only as good as its method, and a credible estimate shows how it was made. The common methods are estimation by analogy (comparing with similar systems built before), bottom-up estimation (breaking the scope into tasks and estimating each), and functional size measurement, such as function points, which counts what the system does for users regardless of how it is built. Good estimates give a range, list their assumptions and name the parts with the most uncertainty, usually integrations and data migration. Be wary of a single figure produced within hours of a one-paragraph description.

A simple way to make the uncertainty visible is a three-point estimate: for each part of the scope, an optimistic, a most likely and a pessimistic number of hours. A common weighting, from the PERT method, is (optimistic + 4 x most likely + pessimistic) / 6, and the spread between the optimistic and pessimistic figures shows where the risk sits. Software development cost estimation also improves as the work proceeds: an estimate made after a short discovery phase, with the main flows, data and integrations written down, is far more reliable than one made from an idea, which is why we quote a fixed price only once the scope is in writing.

How to compare quotes

  • Is it a fixed price for a written scope, or an estimate billed by the hour? (A software requirements specification is the best written scope.)
  • Does it include discovery, design, integrations, data migration and testing, or only building?
  • Which integrations are included, and what is assumed about the other systems' interfaces?
  • How much data is migrated, and who cleans it?
  • Does it include hosting set-up, security, backups and monitoring?
  • What is excluded, in writing?
  • How long are faults fixed at no extra charge after launch?
  • What does maintenance cover, and what does running the system cost each month?
  • Is the source code, the hosting and every account in your company's name?
  • Who actually does the work, and can you see systems they have built?

Ways to spend less without cutting quality

  • Fix the scope before the price. A short discovery phase is the cheapest way to avoid paying for the wrong system.
  • Launch the core first. The workflows used every day pay for the system; the rest waits for evidence.
  • Buy the commodity parts. Sign-in, payments, email, document storage and analytics are solved problems.
  • Migrate less. Move the data you use; archive the rest in a readable format.
  • Own the code. Code in your own repository keeps every future change open to competition.

Questions people ask

How much does custom software development cost?

It depends almost entirely on scope: the number of user roles and workflows, the integrations with other systems, the data to migrate, the security and compliance requirements, and how the system is run after launch. Two systems that sound similar in one sentence can differ several times over in work. Any figure quoted without a written scope is a guess.

Why are custom software quotes so different from each other?

Because they usually describe different systems and different contracts. One quote includes discovery, design, integrations, data migration, testing and hosting set-up; another covers only building screens from a list. Rates also follow local labour costs, which in the EU ranged from 12.0 to 56.8 euros an hour across the whole economy in 2025, according to Eurostat. Compare quotes against the same written scope.

Is custom software more expensive than buying SaaS?

At the start, almost always. Over several years, not necessarily. A subscription is paid per user every year for as long as you use it, and you adapt your process to the product. Custom software costs more up front, then costs maintenance and hosting, and fits your process exactly. Compare the total over five years for the same users and scope, including the cost of the workarounds a product forces.

How much does custom software cost to maintain?

Plan for maintenance as a recurring cost from the start. Software needs security updates, dependency upgrades, fixes, small changes as the business changes, monitoring and hosting. A written maintenance scope agreed before launch turns that into a predictable line in the budget instead of a series of surprises.

Fixed price or time and materials: which is cheaper?

Neither by default. A fixed price suits a scope that can be written down and is the safer choice for the buyer when it can. Time and materials suits work whose scope can only be found by doing it, such as discovery or a product still looking for its users. Many projects combine them: a fixed-price first phase, then agreed phases priced one at a time.

Sources

Your scope checklist

Before you request quotes, write one line for each of these. The answers make quotes comparable and usually shorten the work.

  • The problem the system solves, and how you will know it is solved.
  • Who uses it: each role, roughly how many people, and where.
  • The main workflows, as a short list of steps from start to finish.
  • The systems it must connect to, and whether they have documented interfaces.
  • The data to migrate, from where, and how much history.
  • Reports and exports needed in the first version.
  • Sensitive data and the rules that apply to it.
  • When it must be available: office hours or around the clock.
  • Interfaces: web, mobile, offline use, languages.
  • Timeline, including any fixed date.
  • Maintenance after launch, and who supports users.
  • Ownership: code, hosting and accounts in your company's name.